数据开采中的知识表示和知识发现方法研究

Based on the research of cloud models and uncertainty reasoning technology, the method of knowledge representation is established and algorithms for backward cloud generators are implemented. Therefore transforms between qualitative concepts and their quantitative expressions become much easier and interchangeable by using the given algorithms of forward and back-forward cloud generators respectively. With natural language values and rules represented by cloud models, association algorithms and classification algorithms were put forward and successfully applied to the mining on geographical spatial databases. A novel method called data field, which extended the physics field to the discovery state space, was a breakthrough in the cognitive processing. We tried to utilize data field concept to simulate the cognition and thinking of human being. Furthemore, the method of data field was successfully applied to characteristics selection, pattern recognition, and data mining visualization. The research opened out the mechanism, methods and a set of new tools for knowledge discovery, and provided a new theory for research of artifical intelligence.

建立用于定性定量之间映射的、基于云模型的知识表示方法，提高发现知识的鲁棒性；深化和完善发现状态空间理论，给出能在多个抽象层次上发现广义、特征、关联、序列等不同类型知识的机理和方法；开发一个基于实际应用的知识发现系统原型，同时研究从数据库中发现宏知识的方法。研究成果对建立各类数据开采系统、辅助决策系统有普遍意义。